how to compromise the best price? a group tacit knowledge
TRANSCRIPT
How to compromise the best price?
A group tacit knowledge-based multicriteria ap-proach
Pierre-Emmanuel Arduin, Brice Mayag, Elsa Negre, Camille
Rosenthal-Sabroux
LAMSADE, Universite Paris-Dauphine, France
How to compromise the best price?A group tacit knowledge-based multicriteriaapproach
Tacit knowledge plays an important role in decision making that is why, inthis paper, we propose a group tacit knowledge-based multicriteria approachfor decision process. This approach is based on a practical case study in thefield of second hand vehicle purchase. This paper proposes ideas and researchissues on how to tackle tacit knowledge in group decision processes with cost-benefit and multicriteria decision analysis. Because tacit knowledge cannotalways be explicited, we use the multicriteria decision analysis to fittinglyexplicit tacit knowledge which can be explicited. In order to respect the fi-nancial policy of the company that purchases/sales second hand vehicles, wecontinue using the cost-benefit analysis already implemented. The goal is tohelp the buyer and the seller to compromise the best price.
Keywords: tacit knowledge; decision process; cost-benefit analysis; multi-criteria decision analysis; group decision.
1 Introduction
We know that tacit knowledge plays an important role in decision making
that is why, in this paper, we will discuss a new way to take into account
tacit knowledge for decision-making process. Our approach is based on a
practical case study in the field of second hand vehicle purchase. To the
best of our knowledge, in the literature, no research addresses this kind of
approach. Based on our previous work (Arduin et al., 2013b), this paper
proposes ideas and research issues on how to tackle tacit knowledge for
decision process with cost-benefit and multicriteria analysis based on a
real-world compromise case.
In this paper we present the (group) decision making process, then we
emphasize tacit knowledge in the decision making process. Our work is
based on three postulates, which are (i) knowledge is not an object, (ii)
knowledge is linked to action, and (iii) within a company there exists two
main knowledge categories. We also present the industrial context of our
study: purchase and sale of second-hand vehicles. Actually in the com-
pany they have already defined a decision process to sell a car but tacit
knowledge is not taken into account.
We propose to combine multicriteria analysis (Roy, 1996), particularly
ELECTRE TRI method (Figueira et al., 2005) and cost-benefit analysis
(Nas, 1996). Because tacit knowledge cannot always be explicited, we use
the multicriteria analysis to fittingly explicit tacit knowledge which can
be explicited. In order to respect the financial policy of the company, we
use the cost-benefit analysis. So it is possible to make the best decision
regarding the purchase price according to the different available criteria.
Thus we help the company that purchases/sales second hand vehicles, to
realize the best deal in this compromise context of group decision.
Note that we started from ground observations and that the approach
proposed in this article is specific to the case that we met. Note also that,
to the best of our knowledge, there exists no formal modeling, especially
with multicriteria analysis, of tacit knowledge. Finally, we are aware that
multicriteria analysis can be used alone (without combination with cost-
benefit analysis), but because cost-benefit analysis is already implemented
and used by the society, we propose an approach combining multicriteria
analysis and cost-benefit analysis.
This paper is organized as follows: Section 2 introduces tacit and ex-
plicit knowledge and decision process. Section 3 presents the context of
our real-world case while, in section 4, we describe the existing problem
in such a case. Section 5 explicits how to elaborate a tacit knowledge
decision model to take into account tacit knowledge in decision process.
Finally, Section 6 concludes this paper and provides future work.
2 Tacit and explicit knowledge and decision process
2.1 The company’s knowledge
The company’s knowledge consists of tangible elements (databases, pro-
cedures, drawings, models, algorithms, documents used for analyzing and
synthesizing data) and intangible elements (people’s abilities, professional
knack, “trade secrets”, “routines” - unwritten rules of individual and col-
lective behavior patterns (Nelson and Winter, 1982) -, knowledge of the
company’s history and decision-making contexts, knowledge of the com-
pany environment (clients, competitors, technologies, influential socio-
economic factors)) (Grundstein et al., 2003). They characterize a company
capability to design, produce, sell, support its products and services. They
are representative of the company’s experience and culture (Davenport and
Prusak, 1998). They constitute and produce the added-value of its organi-
zational and production business processes.
Tangible elements are “explicit knowledge”. Heterogeneous, incom-
plete or redundant, they are often marked by the circumstances under which
knowledge was created. They do not express the unwritten rules of those
who formalized knowledge, the “unspoken words”. They are stored and
disseminated in archives, cabinets, and databases.
Intangible elements are “tacit knowledge”. Acquired through practice,
they are adaptable to the situations. Explicitable or non-explicitable, they
are often transmitted by implicit collective apprenticeship or by a master-
apprentice relationship. They are located in people’s minds. Here we are
referring to the knowledge classification of Polanyi (1967). He classifies
the human knowledge into two categories: tacit knowledge and explicit
knowledge. “Tacit knowledge is personal, context-specific, and therefore
hard to formalize and communicate. Explicit or ’codified’ knowledge, on
the other hand, refers to knowledge that is transmittable in formal, sys-
tematic language” (Polanyi, 1967). Our point of view can be found in the
work of Nonaka and Takeuchi (1995), with reference to (Polanyi, 1967),
consider that “tacit knowledge and explicit knowledge are not totally sep-
arated but mutually complementary entities.” For Nonaka and Takeuchi
(1995), explicit knowledge can be easily expressed in written documents
but is less likely to result in major decisions than tacit knowledge, which
is to say that the decision process stems from knowledge acquired through
experience, albeit difficult to express in words. These observations con-
cerning knowledge in the company context highlight the importance of
tacit knowledge. They point out the interest in taking into account tacit
knowledge in decision process.
In the following paragraphs, we will first present the general princi-
ples that underlie decision support research, and especially the formalized
decision-making process by Simon (1977).
2.2 The decision-making process
A process is a combination of phenomena, conceived to be active and or-
ganized in time. A process is a sequence of activities that produces a result
of observable value in UML terms. It can be expressed as different di-
agram (sequence, collaboration or activity). As such, a decision-making
process comprises everything that takes place in reality, like actions, activ-
ities and phenomena that lead to the choice of the final decision. According
to (Simon, 1977) a decision-making process passes through four different
phases: the “Investigation” (Intelligence) phase, the “Design” phase, the
“Choice” phase, and the “Review” phase. With respect to the process de-
fined in (Simon, 1977), an additional important aspect, which underlies
all phases, is the communication and interaction dimension. The empha-
sis on communication and interaction means fostering a minimum on tacit
knowledge. Today, with Information and Communication Technologies
(ICT), thanks to social network and others software, people can share tacit
knowledge. In this respect, the “Investigation” (Intelligence) phase, be-
ing the first phase in Simon’s decision-making process (Simon, 1977), is
primordial. Human are not technology and the decision process cannot be
design as a program that individual can follow precisely and without think-
ing. It leads to the tacit knowledge of people. The relevance of the entire
process depends on this first phase.
It is important to remark that, within a company, decisions can be made
by individuals. However, most of the decisions come from a working group
which can contribute or make a group decision. (Fulop, 2005) indicates
that “Group decision is usually understood as aggregating different indi-
vidual preferences on a given set of alternatives to a single collective pref-
erence. It is assumed that the individuals participating in making a group
decision face the same common problem and are all interested in finding
a solution. A group decision situation involves multiple actors (decision
makers), each with different skills, experience and knowledge relating to
different aspects (criteria) of the problem. In a correct method for syn-
thesizing group decisions, the competence of the different actors to the
different professional fields has also to be taken into account.”
2.3 Tacit knowledge and the decision process
When (Polanyi, 1967) introduces the concepts of sense-giving and sense-
reading, we simply observe that we continuously appropriate information
which is not ours. He defined them as follows: “Both the way we endow
our own utterance with meaning and our attribution of meaning to the ut-
Figure 1: Tacit knowledge transfer
terances of others are acts of tacit knowing. They represent sense-giving
and sense-reading within the structure of tacit knowing” (Polanyi, 1967).
As the authors of this paper, we have got tacit knowledge that we have
structured into information during a process of sense-giving. As the read-
ers of this paper, you have interpreted this information perceiving forms
and colors, integrated words, data, during a process of sense-reading pos-
sibly creating new tacit knowledge for you (see Figure 1).
When a person P1 structures its tacit knowledge and externalizes it, he
creates information. A person P2 perceiving some data from this infor-
mation and internalizing it, possibly creates new tacit knowledge. Thus
knowledge is the result of the interpretation by someone of information
(Arduin et al., 2013a). This interpretation is done through an interpretative
framework that filters data contained in the information and the use of pre-
vious tacit knowledge as presented by Tsuchiya (1993) and by Grundstein
(2012).
Our work is based on three postulates:
Postulate 1: Knowledge is not an object1
Generally, knowledge cannot be treated as an object. It is the result of the
encounter of data and a subject and it is processed within the interpretative1We consider, according to the dictionary, that an object is anything visible or tangible; a material
product or substance.
scheme of the subject’s memory. This postulate is based on the theories
developed by S. Tsuchiya, who deals with the construction of tacit indi-
vidual knowledge (Tsuchiya, 1993). According to his research, the tacit
knowledge which resides within our brain is the result of the meaning we
attribute - through our interpretative schemes - to the data that we perceive
as part of the information that we receive. This individual knowledge is
tacit and it can or cannot be expressed.
Postulate 2: Knowledge is linked to action
From a business perspective, knowledge is created through action. Knowl-
edge is essential for the business’s functioning and is finalized through its
actions. Hence, one has to be interested in the activities of the actors -
decision-makers, engaged in the process contained in the company’s mis-
sion. This vantage point is included in the use of the concept of knowledge,
which cannot be detached from the individual placed within the company,
his/her actions, decisions and relations with the surrounding systems (peo-
ple and artifacts).
Postulate 3: Within a company, there exist two main knowledge categories
Within a company, knowledge consists of, on the one hand, explicit knowl-
edge comprising all tangible elements and, on the other hand, tacit knowl-
edge, which comprises intangible knowledge. Hence, knowledge within
a business organization comprises two main categories: formalized and
explicit knowledge, which can be called “business know-how”, and the
individual and collective tacit knowledge, which can be called the “busi-
ness skills”, and which can either be made explicit or not, which can be
explicitable or not.
3 Purchase/sale of second-hand vehicles: How does it work?
The context of this work has already been presented in (Arduin et al.,
2013b).
3.1 The idea of buying a second-hand vehicle
The company we are interested here, offers a purchase service of second-
hand vehicles without conditions. Indeed, to sell a vehicle to a distributor,
this vehicle must be in good condition and not too old. More generally,
in France, a distributor buys your car only if you buy a new one in his
company. He can afford to offer a higher purchase price because he will
make a significant profit margin with the car you will buy. The promise of
this company is to buy all cars, regardless of their condition and without
obligation to buy an other vehicle.
3.2 Sell a car, how does it work?
When a private individual, Elsa, wants to sell her vehicle, she asks a prior
valuation of the second-hand vehicle on the company website. She gives
information about the main characteristics of the vehicle such as brand,
model and age. Following this query, she obtains a valuation of the pur-
chase price of the vehicle. The system does not take into account the wear
on the vehicle and it believes that the vehicle is in good condition. When
querying to obtain a first valuation on the website, Elsa leaves her address
and telephone number, she will be contacted by the company to have an
appointment in an agency of the company. When Elsa goes to the agency,
she is received by a vehicle expert, John. He examines the vehicle and
values the costs to recondition the vehicle. This step is very important be-
cause it controls the working state of the vehicle, John must do a road test
and identify anomalies. The data are inputed and stored through a com-
puter platform in order to ask the quotation experts the vehicle valuation.
The value that John will get from the purchasing department through this
platform will be the best proposal and John will not have scope on this
proposal.
4 How to take into account tacit knowledge?
How vehicles are evaluated is not discussed here. However, this valuation
is unpredictable because it depends on each quotation expert. To better
estimate the market, these experts go on different ad sites selling vehicles
on the web and examine prices. They valuate the vehicle so that it is well
placed among similar ads. This approach sometimes leads to the purchase
price offered to the private individual is not consistent with the real resale
price.
The profitability of the purchase/sale business is based on one principle:
buy second-hand cars and sell them more expensive, with a profit margin
nearly constant. The issue of valuating vehicles is to know how much the
vehicle may be sold. Therefore one has to calculate a purchase price, re-
moving the wanted profit margin. Nowadays, quotation expert valuations
do not always correspond to the reality of the market. Sometimes the com-
pany does not break even: the purchase price is too high compared to the
final resale price or profit margins are too large while the company could
offer a better price. The company studied the margins. This study shows
that the wanted profit margin is achieved in less than half of the vehicles.
These errors, whether against or in favor of the private individual, de-
crease the quality and the image of the company and make the activity
uncertain. Thus, the business management is difficult to predict and the
behavior of the quotation experts changes. To protect themselves from
a possible market decline, the quotation experts propose purchase prices
lower than they should. Private individuals are disadvantaged and it cre-
ates dissatisfaction and discontent among the private individuals who come
to sale their vehicle.
Taking into account tacit knowledge would improve the vehicle price
valuation: the context of a particular sale, the repair costs, the location of
the vehicle and the market.
Note that data corresponding to the concerned vehicles are stored in a
database (in the next section, vehicles are called actions in the multicrite-
ria analysis). Furthermore, the following knowledge have been obtained
through interviews and exchanges with experts and sellers.
The context of a particular sale. For example, financial or family prob-
lems can cause an urgent need to sell the vehicle regardless of the price or
otherwise can cause an attempt to negotiate the price as high as possible.
The context of the sale could have a strong influence on the final price and
that is the reason why it has to be considered by our proposition. This con-
text is important for the seller and impacts his final decision. We have three
cases: the owner is in no hurry to sell his car (T1), the owner is moderately
hurry to sell his car (T2) or the owner is in a hurry to sell his car (T3). More
the context is supported, more the rating will be high. Here, T1 is better
than T2 and T3.
The repair costs. For example, a simple scratch can cause the replacement
of an important part of the vehicle and only an expert can know that. Repair
costs can be difficult to evaluate notably when the car cannot be studied in
detail. Our proposition gives to John, the vehicle expert, the opportunity
to say rapidly if and how much according to him the repair costs influence
the vehicle final price, although the car is not studied in detail. John can
be able to say, for example, if a simple scratch will cause the replacement
of an important part of the vehicle body. These repair costs impact John
and the company final decision. We have four cases: any repairs (just
polished) (R1), some repairs (single scratch) (R2), a certain number of
repairs (some costs required) (R3) or many repairs (R4). More the repair
costs are supported, more the rating will be high. Here, R1 is better than
R2, R3 and R4.
The location of the vehicle. For example, the experts know that some ve-
hicles will be sold differently depending on their geographical location.
The location impacts John (the vehicle expert) and the company final de-
cision. We have three cases: the vehicle is located in a geographic area
where vehicles are sold at low prices (L1), at medium prices (L2), at high
prices (L3). More the location is supported, more the rating will be high.
Here, L3 is better than L2 and L1.
The market itself. For example, if a vehicle is very popular, the company
will keep it in stock less time, so it will be cheaper. However, if a vehicle
is not popular, it will spend a longer time between the purchase of the ve-
hicle and its resale, so there is an additional storage cost for the company.
The market impacts John (the vehicle expert) and the company final deci-
sion. We have four cases: the vehicle is not popular (M1), the vehicle is
moderately popular (M2), the vehicle is popular (M3) or the vehicle is very
popular (M4). More the market is supported, more the rating will be high.
Here, M4 is better than M3, M2 and M1.
Note that we can speak about group decision because the final com-
promise decision has to take into account knowledge that affects the seller
(Elsa) decision and knowledge that affects John and the company decision.
5 Our proposition
The context presentation shows that John, the vehicle expert who receives
the private individual seller Elsa, has no leeway on the proposed purchase
price. However, it would be interesting to give him a minimum of oppor-
tunities to try to reconcile the purchase price and the sale price wanted by
Elsa. Indeed, sometimes, because of a few euros, a sale may or may not
occur. However, John must also stay close to the expectations of company,
namely, “business is business” and because the company will not accept
losing money.
Our proposal is to develop a tool for risk analysis in terms of decision
making. In such a way that this tool would allow John to slightly higher
modify the proposed purchase price while ensuring that the company will
still be able to make his profit margin and especially not to loose money. It
is therefore to take into account the various available indicators (relational
context, emergency of the sale for the private individual, ...) to cross these
indicators with the vehicle-related criteria (the market, repairs, ...) and test
the impact on the financial expectations of the company. More specifically,
it is to combine the cost-benefit analysis (Nas, 1996) to respect the financial
policy of the company and the multicriteria decision analysis (Roy, 1996),
so that John makes the best decision regarding the purchase price according
to the different available criteria.
5.1 Cost-Benefit Analysis
Cost-benefit analysis is a process that calculates and compares benefits and
costs of a project (see (Nas, 1996) for more details). The main purpose is
to compare the total expected cost of each option against the total expected
benefits. Note that benefits and costs are expressed in monetary terms. It
helps predict whether the benefits outweigh its costs, and by how much
relative to other alternatives.
(Argyrous, 2010) breaks down the cost-benefit analysis into 10 steps:
(1) Determine if the cost-benefit analysis worth doing, (2) Identify objec-
tives and policy alternatives, (3) Determine who has standing, (4) Identify
the costs and benefits of each alternative, (5) Sort into measurable and non-
measurable costs and benefits, (6) Measure costs and benefits that can be
measured in money terms, (7) Conduct sensitivity analysis, (8) Compare
costs-benefits across alternatives, (9) Adjust for non-measurable costs and
benefits and (10) Make a decision.
The cost-benefit analysis recognizes that each choice has a cost and
makes explicit hidden costs and benefits. As results, cost-benefit analysis
returns some indicators such as a ratio of costs divided by benefits. Un-
fortunately, the valuation of life is difficult or impossible and this kind of
approach is ignorant of the political context of decision-making.
5.2 Multicriteria decision analysis
Multicriteria decision analysis (MCDA) is the activity which provides a
decision maker (DM) with a prescription on a set of decision alternatives
(options), when facing multiple, usually conflicting points of view also
called criteria. In fact, there does not exist a unique optimal solution for
such problems and it is necessary to use decision maker’s preferences to
differentiate solutions. For instance, Table 1 presents a set of seven alter-
natives (207 Peugeot cars) evaluated on the four criteria Context, Repair
Cost, Location and Market.
Usually, three types of problems are put forward in the MCDA context
(Roy, 1996, Figueira et al., 2005):
• the choice problem which aims to recommend a subset of alternatives,
as restricted as possible, containing the “satisfactory” ones;
Context Repair Cost Location Marketa1: 207 1.6 HDI Trendy 3p T1 R1 L1 M2a2: 207 1.4 HDI Premium 5p T3 R2 L1 M2a3: 207 1.4 HDI FAP 5p T2 R2 L2 M2a4: 207 1.6 HDI Sport 3p T1 R1 L1 M3a5: 207 1.6 HDI Executive 5p T3 R2 L1 M1a6: 207 1.6 HDI Sport 3p T1 R4 L3 M3a7: 207 1.4 HDI Urban Move 3p T3 R1 L2 M2
Table 1: A performance matrix of selling 207 Peugeot cars
• the sorting problem which aims to assign each alternative into pre-
defined categories or classes;
• the ranking problem which aims to order the alternatives by decreas-
ing order of preferences.
(Argyrous, 2010) breaks down the multicriteria decision analysis into 7
steps: (1) Establish the decision context, (2) Identify the value/performance
criteria, (3) Describe/rate the performance of each option against the crite-
ria, (4) Assign weights across criteria, (5) Combine the scores and weights,
(6) Examine the results and review and (7) Conduct sensitivity analysis.
The multicriteria decision analysis provides an audit trail, especially
in situations where decision-making is required to follow rules and to be
justified in explicit terms and is based on weightings. Because weights
are in general difficult to obtain in a direct discussion with the DM, some
methods, like ELECTRE TRI, propose to elicitate them in indirect way. To
do this, these methodologies ask to the DM some preferences over the set
of alternatives.
ELECTRE TRI method
ELECTRE TRI (Figueira et al., 2005) is a MCDA method which deals
with the ordinal (“qualitative”) sorting problematic. Because our aim, in
this section, is to show that the aggregation of tacit knowledge through a
MCDA method is possible, we present hereafter a simple version of ELEC-
TRE TRI without elicitation of preference thresholds and veto. Therefore
this version is sufficient in our context. The preference thresholds and the
vetoes can be added and elicitated by the practician, in interaction with the
DM, when he faced with a more complex decision problem.
Let us denote by A = {a1; a2; . . . ; am} a set of m alternatives or op-
tions, N = {1; 2; . . . ;n} a set of n criteria, C = {C1;C2; . . . ;Ct} a set
of ordered categories (C1 is the worst one and Ct is the best one) and
B = {b0; b1; . . . ; bt} a set of profiles (reference alternatives) that separate
consecutive categories. Each category Ci is limited by two profiles: bi is
the upper limit and bi−1 is the lower limit.
The assignment of alternatives to categories is based on the concept of
outranking relation on A × B. An alternative ai ∈ A outranks a profile
bh ∈ B (denoted ai S bh) if it can be considered at least as good as the
latter (i.e., ai is not worse than bh), given the evaluations (performances)
of ai and bh at the n criteria. If ai is not worse than bh in every criterion,
then it is obvious that ai S bh. However, if there are some criteria where aiis worse than bh, then ai may outrank bh or not, depending on the relative
importance of those criteria and the differences in the evaluations (small
differences might be ignored). Roughly speaking,
ai outranks bh(ai S bh)⇔n∑1
kj cj(ai, bh) ≥ λ.
Where
• cj(ai, bh) =
{1 if ai %j bh
0 otherwise.
The relation ai %j bh means that the performance of ai on the crite-
rion j is at least as good as the performance of bh on the same criterion
j.
• kj is the importance (weight) of criterion j such thatn∑1
kj = 1.
• λ is the cutting level i.e. a threshold that indicates whether the credi-
bility is significant or not. This parameter is often taken between 0.5
and 1.
Hence ELECTRE TRI assigns the alternative ai to the highest category
Ch such that ai outranks bh−1 i.e.
ai belongs to category Ch ⇔ ai S bh−1 and not(ai S bh)
5.3 Elaboration of a tacit knowledge decision model
Because cost-benefit analysis is already implemented and used by the so-
ciety, in this section, we focus on MCDA, in order to concentrate on non-
measurable costs and benefits.
Thus, in this section we show how to construct a decision model from
the tacit knowledge embodied in experts and John’s mind i.e. how they will
be confronted in order to decide if and how much the final price could be
increased by John, the vehicle expert. Given a vehicle evaluated on Con-
text, Repair Cost, Location and Market, the aim of John here is to assign
it to one of the following six pre-defined categories or classes: no increase
(C1), increase at most 5% of the vehicle estimated price (C2), increase at
most 10% of the vehicle estimated price (C3), increase at most 15% of
the vehicle estimated price (C4), increase at most 20% of the vehicle esti-
mated price (C5), and increase at most 25% of the vehicle estimated price
(C6). These parameters, as well as most of the parameters introduced in
this study, have been defined by the considered organization and its finan-
cial policy. Moreover, relying on previous sales and on his expertise, John,
the vehicle expert, is able to value the influence of tacit knowledge on a
vehicle final price.
Context Repair Cost Location Marketb0 T3 R4 L1 M1b1 T3 R3 L1 M1b2 T3 R3 L1 M2b3 T2 R2 L2 M2b4 T2 R2 L2 M3b5 T1 R1 L3 M3b6 T1 R1 L3 M4
Table 2: Profiles of the categories defined by John
John’s problem can be viewed as a MCDA sorting problem. Therefore
we solve it by using the software IRIS which implements the ELECTRE
TRI method seen above. IRIS is available at http://www.lamsade.
dauphine.fr/spip.php?rubrique64 and requires as the input the
set of profiles B associated to the six categories C1, . . . , C6. In Table 2, we
have the seven profiles given by John. According to these profiles, the
assignments of vehicles of Table 1 and the parameters of the model con-
structed are given in Figure 2.
According to these profiles, the assignments of vehicles of Table 1 and
the parameters of the model constructed are given in Figure 2. For instance,
one can see that the weight associated to the criterion “Context” is 0.1825,
the value of the cutting level λ is 0.5825 and the categories assigned to a1and a7 are respectively C4 and C5.
Given the preferential information given by John, the results obtained
by applying ELECTRE TRI of IRIS show that:
• There is no vehicle assigned to the three categories No increase (C1),
Increase at most 5% of the vehicle estimated price (C2) and Increase
at most 25% of the vehicle estimated price (C6).
• The vehicles a2 (207 1.4 HDI Premium 5p) and a5 (207 1.6 HDI Ex-
ecutive 5p) are assigned to the category Increase at most 10% of the
vehicle estimated price (C3).
Figure 2: A MCDA tactit knowledge model computed by IRIS
• The cars a1 (207 1.6 HDI Trendy 3p) and a3 (207 1.4 HDI FAP 5p) are
assigned to the category Increase at most 15% of the vehicle estimated
price (C4).
• The last three vehicles a4 (207 1.6 HDI Sport 3p), a6 (207 1.6 HDI
Sport 3p) and a7 (207 1.4 HDI Urban Move 3p) are assigned to the
category Increase at most 20% of the vehicle estimated price (C5).
• The weights associated to the criteria “Context”, “Repair Cost”, “Lo-
cation” and “Market” are respectively 0.1825, 0.3525, 0.2825 and
0.1825. Hence “Repair Cost” appears as the most important criterion.
• The value of the cutting level λ used is 0.5825.
As a reminder, Elsa wants to sell her car to the company that employs
John. She wants to obtain the highest possible price. John, without preju-
dicing Elsa, must propose a reasonable price in line with the policy of the
company that sold later the Elsa’s vehicle to another individual. The pur-
chase price of the Elsa’s vehicle is first proposed by the quotation experts
and is based on the technical characteristics of the vehicle.
Hitherto John was limited to just offer the price given by the quotation
experts. Our approach allows to give some leeway to John on the price
given by the experts.
In our context, Elsa is the only decision maker. Indeed, following the
proposal of John, Elsa has the ability to accept or refuse to sell her vehicle
to the company of John. However, the process of compromise on the price
of the vehicle is a process of group decision making since it takes into
account the “preferences” of:
• John, who is representing the interests of the quotation experts and
the company,
• the quotation experts who have the technical expertise,
• the society, for whom “business is business”,
• Elsa who wishes to obtain the highest possible price and do not want
to feel prejudiced.
Therefore, the fact that no vehicle is assigned to C1 (no increase) and
C2 (increase at most 5%) indicates that Elsa should not feel prejudiced
because John will be able to raise the price. Similarly, the fact that no ve-
hicle is assigned to C6 (increase at most 25%) indicates that the company
does not take much financial risk and its employee, John, has a limited
flexibility. Thus, the fact that the categories C3 (increase at most 10%),
C4 (increase at most 15%) and C5 (increase at most 20%) are the most
probable indicates that the final decision is a good compromise (a decision
group), which takes into account all the “preferences” of all stakeholders.
In reality, the vehicle a2 (207 1.4 HDI Premium 5p) was quoted 9890
euros by the quotation experts. John, before our proposal, had no flexibil-
ity on the price. Thus, when he proposed this price to Elsa, she refused.
Indeed, in France in particular, there are levels values. In fact, if John
had proposed a price slightly higher than 10000 euros, Elsa would have ac-
cepted. Finally, thanks to our approach, in a similar situation, John now has
a leeway of at most 10% (see previously, a2 is assigned toC3), i.e., John can
offer prices belonging to the bounded interval [ 9890, 10879 [(containing,
for example, the level value 10000 euros) and Elsa will agree to sell her
vehicle to the company of John.
It should be noticed that John’s problem can be also solved by any
MCDA classification or sorting methods such as Dominance-based rough
set approach (DRSA) (Greco et al., 2001). In this case, because each
MCDA sorting method lead to a decision model different to another method
(they do not have the same inputs and the same parameters), it will not be
surprising to have a different result compared to what we have obtained
above.
5.4 Combining cost-benefit and the tacit knowledge model
Even though, (Dobes and Bennett, 2009) seems to interpret that the mul-
ticriteria decision approach goal is worst than the cost-benefit approach
goal, cost-benefit analysis and multicriteria decision analysis share some
similarities:
• both are frameworks for assessing options facing decision-makers,
• both try to construct a metric for comparing options,
• both are sensitive to assumptions, but these assumptions are different.
Remember that cost-benefit analysis is already implemented and used
by the society and MCDA is efficient for non-measurable costs and bene-
fits. The society wants only one system. So, we have to proceed into two
steps / analyses and then, to combine them.
We argue to benefit of the advantages of each kind of analysis. The
cost-benefit analysis is pertinent and easy-to-use for measurable costs and
takes into account the expectations of the company (“business is busi-
ness”). The multicriteria decision analysis is pertinent to make decision
and allows to bring into play some “non-measurable” costs. Thus, our
proposition consists in using the cost-benefit analysis for the measurable
costs (quantitative ones) and using the multicriteria decision analysis for
the “non-measurable” costs (qualitative ones), i.e. the tacit knowledge that
can affect the purchase price. The combination of these two approaches
can be done by using a MCDA approach with two criteria:
• The first criterion will concern the cost-benefit;
• The second criterion will be an aggregation of the tacit knowledge, as
done above using the ELECTRE TRI method, where the evaluation of
a vehicle on this criterion will be one of the six pre-defined categories
C1 to C6.
This combined approach is presented in Figure 3 as a hierarchy of criteria.
Because data concerning cost-benefit were not provided by John’s com-
pany, we tried to model only, in this hierarchy, the node associated to tacit
knowledge.
6 Conclusion and future work
In this paper we emphasize the strengths of tacit knowledge in the deci-
sion making process in addition to decision support systems. Thus, we
elaborated the MCDA model based on four criteria (context, repair costs,
location, market). In order to respect the financial policy of the company,
we continue using the cost-benefit analysis already implemented in the
field. Our MCDA model plays an important role in refining the vehicle
final price, so that not only the financial expectations of the company are
Figure 3: A hierarchy MCDA model as an aggregation of cost-benefit and tacit knowledge
taken into account, but also the final vehicle price satisfies the seller and
ensures that the transaction will be done.
In the future, our works will focus on identifying and formalizing the
explicitable part of tacit knowledge. The materialization of tacit knowl-
edge opens the way towards some new research perspectives, specially for
user-centered design of decision and group decision systems. In addition,
an MCDA model being able to take into account measurable costs, we
are now studying the capability of our MCDA model to integrate not only
non measurable costs, but also those which are considered through a cost-
benefit analysis in the studied field. The company will then hold a new
group tacit knowledge-based decision system covering every parameters
in order to compromise the best price.
References
Arduin, P.-E., Grundstein, M., and Rosenthal-Sabroux, C. (2013a). From
knowledge sharing to collaborative decision making. International Jour-
nal of Information and Decision Sciences, 5(3):295–311.
Arduin, P.-E., Negre, E., and Rosenthal-Sabroux, C. (2013b). Towards a
new tacit knowledge-based approach for decision process. In Proceed-
ings of the first international conference on Knowledge Management,
Information and Knowledge Systems (KMIKS), Tunisia, pages 89–106.
Argyrous, G. (2010). Cost-benefit analysis and multi-criteria analysis:
Competing or complementary approaches? School of Social Sciences
and International Studies, UNSW.
Davenport, T. and Prusak, L. (1998). Working Knowledge: How Organi-
zations Manage What They Know. Harvard University Press.
Dobes, L. and Bennett, J. (2009). Multi-Criteria Analysis: ”Good Enough”
for Government Work?
Figueira, J., Mousseau, V., and Roy, B. (2005). ELECTRE methods. In
Figueira, J., Greco, S., and Ehrgott, M., editors, Multiple Criteria Deci-
sion Analysis: State of the Art Surveys, pages 133–162. Springer.
Fulop, J. (2005). Introduction to Decision Making Methods. Technical
report, Hungarian Academy of Sciences.
Greco, S., Matarazzo, B., and Slowinski, R. (2001). Rough sets theory
for multi-criteria decision analysis. European Journal of Operational
Research, 129(1):1–47.
Grundstein, M. (2012). New Research on Knowledge Management Models
and Methods, chapter Three Postulates That Change Knowledge Man-
agement Paradigm. InTech.
Grundstein, M., Rosenthal-Sabroux, C., and Pachulski, A. (2003). Rein-
forcing decision aid by capitalizing on company’s knowledge: Future
prospects. Open Access publications from Universite Paris-Dauphine
urn:hdl:123456789/4132, Universite Paris-Dauphine.
Nas, T. (1996). Cost-benefit analysis: Theory and application. Sage Pub-
lications, Thousand Oaks.
Nelson, R. and Winter, S. (1982). An Evolutionary Theory of Economic
Change. Belknap Press of Harvard University Press.
Nonaka, I. and Takeuchi, H. (1995). The knowledge-Creating Company.
Oxford University Press.
Polanyi, M. (1967). Sense-giving and sense-reading. Philosophy: Journal
of the Royal Institute of Philosophy, 42(162):301–323.
Roy, B. (1996). Multicriteria Methodology for Decision Aiding. Kluwer
Academic, Dordrecht.
Simon, H. A. (1977). The New Science of Management Decision. Prentice
Hall PTR, Upper Saddle River, NJ, USA.
Tsuchiya, S. (1993). Improving knowledge creation ability through orga-
nizational learning. In ISMICK 1993: Proceedings of the International
Symposium on the Management of Industrial and Corporate Knowledge,
pages 87–95.